Duplex imaging findings predict stenosis after carotid endarterectomy

Duplex imaging findings predict stcnosis after carotid endarterectomy J o n a t h a n Golledge, Rachel C u m i n g , M a r y Ellis, A l u n H . Davies, and R o g e r M. G r e e n h a l g h , London, United Kingdom

Purpose: This study was performed to determine whether early duplex findings predicted restenosis after carotid endarterectomy. Methods: One htmdred ninety-two symptomatic patients who underwent carotid endarterectomy were studied with color duplex imaging at 1 day and 1 week after surgery to identify minor residual disease (causing <50% stenosis), arterial kinking, and suture stricture, and to measure the external and luminal diameters of the carotid bulb and distal internal carotid artery. Patients were then observed prospectively with duplex surveillance for a median of 24 months to identify >50% restenosis. Results: Twenty-five stenoses > 50% of the operated carotid artery (13%) were identified, four at 1 day (residual) and 21 at a median follow-up of 6 months (restenosis). On multiple logistic regression analysis, > 50% restenosis was found to be associated with minor day-1 residual stenosis (p = 0.01) and with small luminal diameter of the distal internal carotid artery (p = 0.03) as measured I week after carotid endarterectomy. Life table analysis showed restenosis at 24 months to be more common for patients with below-median than patients with above-median carotid bulb external diameter (18% vs 5%, respectively; p = 0.01). Conclusions: Duplex scanning within a week of carotid endarterectomy identifies > 50% residual stenosis, in addition to minor residual 25% to 50% stenosis and small carotid dimensions, which are good predictors of >50% restenosis at 6 months. (J Vase Surg 1997;26:43-8.)

Restenosis is identified in 10% to 20% o f patients after carotid endarterectomy when duplex surveillance is performed. 1,2 Systemic factors such as hypercholesterolemia 3 and smoldng 4 have been associated with an increased risk of restenosis; however, the effect of local factors may be more important, s Duplex imaging allows accurate visualization o f the carotid bifurcation after carotid endarterectomy and can identify residual disease or flow disturbances that may stimulate the development ofintimal hyperplasia and restenosis. Flow disturbances may be a particular problem after primary closure o f a small carotid artery when suture stricture may produce a localized flow disturbance, which could predispose the patient to thrombosis and acute stroke in addition to restenosis. 6 In these circumstances, patch repair has been r e c o m m e n d e d because it has been From the Department of Surgery, Chafing Cross and Westminster Medical School. Reprint requests: Jonathan Golledge, 35 Sullivans Reach, Walton on Thame, Surrey, London KT12 2QB. Copyright © 1997 by The Societyfor Vascular Surgeryand International Society for Cardiovascular Surgery, North American Chapter. 0741-5214/97/$5.00 + 0 24/1/79433

shown to increase the diameter of the distal internal carotid artery and carotid bulb. 7 Hence a n u m b e r o f vascular surgeons perform patch angioplasty after performing carotid endarterectomy if the internal carotid artery external diamcter is small, s,9 A n u m b e r of studies, including two randomized studies, have shown the rate ofrestenosis to be lower in patients who receive a patch. 1°,1~ One further randomized study showed restenosis to be more c o m m o n in patched arteries (12% vs 2% in primary closure at 1 year)12; however, in this study patients who had internal carotid arteries less than 5 m m in diameter had obligatory patch repair. Thus there appears to be a complex interaction between a number o f local factors that may influence flow at the endarterectomy site, such as the presence o f residual disease and the diameter of the carotid artery after primary closure or patch repair. We have previously reported on the value o f early duplex scanning in determining the requirement for reexploration in patients in w h o m acute symptoms develop after carotid endarterectomy. 13 In addition, we have demonstrated that restenosis is not predictive o f the later development o f ipsilateral stroke or transient ischemic attack (TIA). 14 In this study we 43

JOURNAL OF VASCULAR SURGERY

44

July1997

Golledge et al.

have investigated the effect of residual disease, flow disturbance, and postrepair carotid dimensions identified in asymptomatic patients by color duplex imaging early after carotid endarterectomy on the later development of restenosis. PATIENTS A N D M E T H O D S One hundred ninety-two patients underwent carotid endarterectomy. The patients' median age was 67 years (range, 38 to 90 years); 134 were men and 58 were women. Thirty-nine patients had a history of amanrosis fugax, 91 had a history of TIA, 11 had a history of transient stroke, 34 had a history of established stroke, 7 had a history of progressive stroke, and 10 had a history of crescendo TIA. Ninety-five patients had a history of hypertension, 72 had a history of peripheral vascular discase, 60 had a history of ischemic heart disease, and 14 had a history of diabetes melfitus. Forty patients were currently smoldng cigarettes. All operations were performed or supervised by the senior author (RMG). The details of the surgical technique have previously been described. 7 Intraoperative calliper measurement of the distal internal carotid artery and carotid bulb were performed. If the distal internal carotid artery measured less than 6 ram, patch repair was performed using a 5 mm lcnitted Dacron patch. Larger carotid arteries were repaired by primary closure. Continuous-wave Doppler scanning was used during surgery to check postrepair flow. The wound was closed with subcuticular prolene, and a clear plastic dressing was placed to allow early duplex assessment. Extensive duplex surveillance was performed after operation (4 MHz probe, Ultramark 9, HDI, Advanced Technology Laboratories, Bothell, Wash.). If any neurologic symptoms developed during the first 24 hours, an urgent duplex scan was undertaken and operative intervention planned on the basis of this scan. The results of this policy have been previously described and have been shown to lower the perioperative stroke rate from 3.4% to 0.8%13; the scanning policy has not been further assessed in this study. Otherwise, the patients were .routinely scanned at 1 day, at 1 week, at 3, 6, 9, and 12 months, and then each year. A detailed description of the outcome of this duplex surveillance has been recently published? 4 In this report we present the findings of the duplex imaging performed within 1 week of operation in relation to the outcome to assess the value of duplex assessment performed at 1 day and at 1 week for patients who were asymptomatic after carotid endarterectomy. Any patient in whom neurologic

symptoms developed underwent urgent duplex imaging, this included five patients who had strokes (2 contralateral and 3 ipsilateral) and l0 who had a TIA within 30 days of operation? 3,14 At the 1-day scan, the internal carotid artery was assessed for kinldng or hemodynamically significant flow disturbance that would be suggestive of narrowing adjacent to the repair site. Kinldng was defined as angulation of the internal carotid artery by more than 50 degrees on the B-mode image. Suture stricture was identified as a narrowing of the full thickness of the wall of the carotid artery that was associated with a hemodynamically significant flow disturbance (peak systolic frequency >3.8 ICHz or peak systolic velocity >1.25 m/s). Residual stenosis was defined as luminal diameter reduction alone identified on B mode and was graded using Doppler frequency shift criteria. A stenosis of 25% to 50% was identified by a peak systolic frequency of 2 to 3.8 ICHz associated with spectral broadening, whereas a stenosis of > 50% was identified by a peak systolic frequency >3.8 KHz? 4 At the 1-week scan, the internal carotid bulb and distal internal carotid artery lumen and external diameters were measured. The reproducibility of duplex ultrasound in the assessment of arterial dimensions has been previously reported? 5 At the subsequent follow-up times, duplex assessment of the ipsilateral carotid artery was performed to detect evidence of >50% restenosis. Thus two types ofstenosis of the operated carotid artery were differentiated, that which was present at 1 day (residual) and that which developed subsequently (restenosis). Data were stored on Dbase 4 (Borland, Scott's Valley, Calif.), and statistical analysis was carried out on Starview (Abacus Concepts, Berkeley, Calif.) using multiple logistic regression and life table analysis to determine the association between patient variables or early duplex findings and the subsequent development of restenosis. Life tables were compared by log-rank test. RESULTS One hundred eleven (58%) of the 192 patients required an intraoperative shunt and 103 (54%) underwent repair with a 5 mm Dacron patch. Patients were followed-up for a median of 24 months (range, 3 to 48 months). The 1-day scan. It was possible to visualize the carotid bifurcation in all 192 patients at 1 day. Kinking of the internal carotid artery was documented in 16 patients (5%), suture stenosis in 20 (10%), >50% residual stenosis in four (2%), asymptomatic occlusion in one, and 25% to 50% residual stenosis in 25.

JOURNAL OF VASCULAR SURGERY Volume 26, Number 1

Golledge et al.

T a b l e I. Characteristics o f patients who developed restenosis

T a b l e I I . Relationship between duplex findings at 1 day and 1 week after carotid endarterectomy and restenosis

No

restenosis

Restenosis

Number 167 21 Median age 67 (38 to 89) 68 (60 to 90) Sex M:F 120:48 11:10 Operation Shunt 96 (57%) 15 (71%) Patch 94 (56%) 9 (43%) History Ischaemic heart disease 52 (31%) 8 (38%) Hypertension 83 (45%) 12 (57%) Diabetes mellitus 13 (8%) 1 (5%) Peripheral vasculardisease 63 (38%) 9 (43%) Smoking 36 (21%) 4 (19%)

45

p

No

restenosis

0.8 0.1 0.1 0.3 0.6 0.8 0.8 0.6 0.6

Significance refersto that calculatedon multiple logistic regression analysis.

Restenosis. Duplex surveillance identified 21 restenoses o f the internal carotid artery at a median time o f 6 months (range, 3 to 24 months). By life table analysis the rate o f >50% stenosis of the operated carotid artery was 19% at 24 months. Tables I and I I list the patient and operation variables and findings identified on the 1-week duplex scan for patients in w h o m restenosis developed in comparison with those who did not. A multiple logistic regression analysis identified only two variables to bc significantly associated with >50% restenosis: minor residual disease (p = 0.01) and small luminal diameter o f the distal internal carotid artery (p = 0.03). Other factors, including gender (p = 0.1) and smoldng history (p = 0.6) were not associated with restenosis. Patching was not associated with restenosis (Table I). Preoperative carotid dimensions, measured by both calliper and ultrasound, did not predict later development of restenosis. Life table analysis demonstrated that minor residual disease identified at 1 day was associated with a subsequent restenosis rate o f 28% by 24 months, which is significantly greater than the 8% restenosis rate for patients in which no residual disease was detected (Fig. 1; p = 0,01). C a r o t i d d i m e n s i o n s a n d restenosis. Figs. 2 and 3 are life table analyses that demonstrate the restenosis rate for patients who had above-median and below-median carotid bulb and distal internal carotid artery diameters, respectively, as measured 1 week after carotid endarterectomy. The restenosis rate was greater for patients who had below-median external (18%; p = 0.02) and luminal (24%; p =

Restenosis

Number 167 21 Kinking 15 (9%) 1 (5%) Suture stricture 18 (11%) 2 (10%) Residual stenosis 19 (11%) 6 (29%) (20% to 45%) Median carotid diameters at 1 week (range): Carotid bulb (external) 8.1 (6 to 14.4) 7.0 (5 to 11) Carotid bulb (luminal) 6.9 (4.2 to 12.4) 5.7 (3 to 9.7) Distal IC (external) 5.9 (3.6 to 8.8) 4.4 (3 to 6.9) Distal IC (luminal) 4.8 (3 to 7.4) 3.4 (2.6 to 5)

p

0.3 0.2 0.01

0.9 0.5 0.3 0.03

Significancerefersto that calculatedon multiple logistic regression analysis. IC, Internal carotid artery.

0.01) carotid bulb and external (19%; p = 0.1) and luminal (24%; p = 0.01) distal internal carotid artery diameters than for patients who had above-median diameters (7%) at 24 months. C a r o t i d d i m e n s i o n s a n d gender. W o m e n generally had smaller carotid arteries before operation. Thus the median distal internal carotid artery external diameter was 5 m m (range, 3 to 8 ram) in w o m e n and 6 m m (range, 4 to 8.5 m m ) in men before operation. Therefore, w o m e n underwent patch repair more often (66%) than did men (50%). As a result, carotid dimensions were more similar after operation: the median distal internal carotid artery external diameter was 5.6 m m (range, 3.2 to 9.1 ram) in w o m e n and 6 m m (range, 4 to 8.8 ram) in men after operation. Despite this similarity, the restenosis rate was greater for women, though not significantly so (17% compared vdth 9% in men at 24 months; p = 0.1). S t r o k e a n d carotid p o s t o p e r a t i v e dimensions. Only three patients had a stJ:oke and 10 a T I A that was related to their ipsilateral carotid artery between 1 and 24 months after operation. A multiple logistic regression analysis failed to identify any association between findings on the 1-day or 1-week duplex scan and subsequent ipsilateral stroke or TIA. This can be explained by the lack o f association between restenosis and ipsilateral stroke and T I A that was previously described. However, the small n u m b e r of patients in w h o m ipsilateral symptoms developed and the limited follow-up need to be borne in mind before maldng conclusions on the importance of the duplex-identified abnormalities.

46

JOURaNAL OF VASCULAR SURGERY July 1997

Golledge et aL

freedom from > 50% restenosis 100

[3-

[3

O"

,,[3

"H

no residual disease (at 1 day)

90

80.

o-------o

¢

.')

1'2

&

2'1

' O 20-50% residual stenosis (at 1 day)

70.

60

t'5

2'4

Time (months) Numbers at risk Time (months) 0 no residual disease 163 20-50% residual stenosis 25

3 155 19

6 140 16

9 12 109 104 13 13

18 83 13

24 67 12

Fig. 1. Development of restenosis in relation to residual disease identified 1 day after carotid endarterectomy.

freedom from >50% restenosis 100t ~

[]

[3 ICBe above median

O'

O

ICBIabove median

O,

O,

ICBebelow median

~

~

ICBI below median

2'1

2'4

90-

~

-¢

80~ 70-

6~ 0

3

6

9 1'2 1'5 Time (months)

1'8

CBe = external diameterof carotid bulb C81 = luminal diameter of carotid bulb

Numbers at risk Time (months) 0 ICBe above median 93 ICBe below medfan 93 ICBI above median 93 [CBI below medfan 93

3 82 81 82 81

6 66 62 68 60

9 64 57 67 55

12 62 57 67 55

1B 50 50 54 50

24 50 45 54 42

Fig. 2. Development ofrestenosis in relation to carotid bulb diameters measured 1 week after carotid endarterectomy.

DISCUSSION Duplex assessment that is performed early after carotid endarterectomy has been shown to be of value in symptomatic patients and its use has also been advocated to identify residual disease and flow abnormalities that should be repaired./6,1r We continue to r e c o m m e n d urgent duplex imaging in patients in w h o m neurologic symptoms develop after

carotid endarterectomy. In this study we have investigated whether duple x findings at 1 day and 1 week after carotid endarterectomy are predictive of restenosis. An association has been demonstrated between minor residual disease associated with a <50% stenosis and the later development of >50% restenosis. Restenosis was also more c o m m o n in small-diameter carotid arteries as demonstrated at 1 week after

JOURNAL OF VASCULAR SURGERY Volume 26, Number 1

Golledge et al.

47

freedom from >50% restenosis

ICDe above median ICDI above median

90-

ICDe below median

ICDI below median

70-

60

l's Time (months) Numbers at risk Time (months) ICDe above median ICDe below median ICDI above median ICDI below median

1'8

2'1

ICDe = outer dimeter of the distal internal carotid artery ICDI = luminal diameter of the distal internal carotid artery

0 93 93 93 93

3 81 81 83 80

6 81 75 80 72

9 64 62 69 61

12 64 60 65 60

18 61 53 63 57

24 49 45 51 50

Fig. 3. Development of restenosis in relation to distal internal carotid diameters measured 1 week after carotid endarterectomy. carotid endarterectomy. For instance, the restenosis rate for patients who had below-median carotid bulb external diameter was 2.5 times greater than that for patients who had above-median diameters. However, from the multiple regression analysis the most important carotid dimension in determining later restenosis appears to be the luminal diameter of the distal carotid artery (Table II, Fig. 3). The diameters o f both the distal internal carotid artery and the carotid bulb are increased by patch repair, 7 thus the apparent effect o f patching on restenosis 1°,12 may reflect to some extent an effect on carotid dimensions. Baker and colleagues, 16 in a study that used intraoperative duplex scanning, demonstrated a similar association between residual disease and the later development of restenosis. They found a 17% restenosis rate for patients who had minor residual disease compared with 4.3% for patients who had a normal duplex scan (p < 0.001). Other studies have also suggested the importance o f residual disease associated with minor degrees ofstenosis in the later development o f more significant restenosis, s38 2o It is believed that restenosis that develops within 1 year of surgery is caused by intimal hyperplasia. 2~ A large number of studies, which principally used ani-

real models, have shown a role for flow (shear stress) in the development o f intimal hyperplasia. 2>24 It could be that flow disturbance that is associated with minor residual stenosis exacerbated by the increased flow velocity that is present in smaller-diameter carotid arteries acts as a stimulus to the development of intimal hyperplasia. On the basis o f this association between residual defects and later restenosis, some authors have advised reexploration and revision of these abnormalities, 16a7 although no study has shown an significant improvement in outcome as a result of such revision. We have avoided revision on the basis o f residual disease because most studies have shown no association betwcen restenosis and ipsilateral stroke or TIA} 6 Our own study demonstrated equal ipsilateral stroke and TIA rates for patients who had >50% restenosis and those who did not. 14 Similarly, most studies have shown no association between abnormalities demonstrated on intraoperative duplex and perioperative carotid artery occlusion. 16 However, with the demonstrated association between minor residual disease and later significant restenosis it could be argued that intraoperative duplex scanning could improve the anatomic outcome. In contrast to the studies discussed above, Kinney

48

JOURNAL OF VASCULARSURGERY July 1997

Golledge et al.

et al. 17 f o u n d patients w h o h a d a b n o r m a l i n t r a o p e r ative D o p p l e r assessment to have an increased ipsilateral stroke rates at 72 m o n t h s after surgery; however, the difference was very small (8% c o m p a r e d w i t h 2%), a n d the n u m b e r s at risk were m i n i m a l by this time (20 a n d 2). A n u m b e r 2s,26 b u t n o t a l p studies have s h o w n restenosis t o be m o r e c o m m o n in w o m e n t h a n in m e n . This difference m a y be r e l a t e d t o t h e i r s m a l l e r c a r o t i d arteries. 16 I n this s t u d y , a l t h o u g h b e f o r e s u r g e r y w o m e n d i d in g e n e r a l have s m a l l e r c a r o t i d arteries, as a r e s u l t o f o u r selective p a t c h i n g p o l i c y a g r e a t e r p r o p o r t i o n u n d e r w e n t p a t c h repair. T h u s b y t h e 1 - d a y scan, c a r o t i d a r t e r y d i a m e ters w e r e similar in m e n a n d w o m e n . D e s p i t e this, r e s t e n o s i s was m o r e c o m m o n in w o m e n (17% c o m p a r e d w i t h 9% in m e n ; p = 0 . 1 ) , s u g g e s t i n g t h a t differences in c a r o t i d d i m e n s i o n s m a y n o t be the only factor involved. CONCLUSION D u p l e x assessment w i t h i n a w e e k o f c a r o t i d enda r t e r e c t o m y identifies significant residual stenosis ( > 5 0 % ) , plus m i n o r residual disease (25% to 50% stenosis) a n d small carotid d i m e n s i o n s predictive o f later restenosis. Given the lack o f association bet w e e n a s y m p t o m a t i c restenosis a n d ipsilateral stroke, we w o u l d n o t r e c o m m e n d reintervention. REFERENCES

1. DeGroote RD, Lynch TG, lamil Z, Hobson RW. Carotid restenosis: long-term non-invasive follow-up after carotid endarterectomy. Stroke 1987;18:1031-6. 2. Mattos HA, van Bemmelen PS, Barkmeier LD, Hodgson KJ, Ramsey DE, Sumner DS. Routine surveillance after carotid endarterectomy: does it affect clinical management? I Vasc Surg 1993;17:819-31. 3. Rapp IH, Qvarfordt P, I@upsldWC, Ehrenfield WK, Stoney RJ. Hypercholesterolaemia and early restenosis after carotid endarterectomy. Surgery 1987;101:277-82. 4. Cuming R, Worrell P, Woolcock NE, Franks PJ, Greenhalgh RM, Powell IT. The influence of smoking and iipids on restenosis after carotid endarterectomy. Eur l Vasc Surg 1993;7:572-6. 5. Bandyk DF, Kaebnick HW, Adams MB, Towne lB. Turbulence occurring after carotid bifurcation endarterectomy: a harbinger of residual and recurrent carotid stenosis. J Vase Surg 1988;7:261-74. 6. Chant ADB. Carotid patch angioplasty. In: Trials and tribulations of vascular surgery. Greenhalgh RM, Fowkes FGR, editors. London: W.B. Sannders, 1996:117-21. 7. Golledge J, Cuming R, Davies AH, Greenhalgh RM. Outcome of selective patching following carotid endarterectomy. Eur l Vase Surg 1996;11:485-9. 8. Hans SS, Girischkumar H, Hans B. Venous patch grafts and carotid eudarterectomy: a critical appraisal. Arch Surg 1987; 122:1134-8. 9. Treiman RL, Foran RF, Wagner WH, Crossman DV, Levin

10.

11.

12.

13.

14.

I5.

16.

17.

PM, Cohen JL. Does routine patch angioplasty after carotid endarterectomy lessen the risk of perioperative stroke? Ann Vase Surg 1993;7:317-9. Eikelboom BC, Ackerstaff RGA, Hoeneveld H, Ludwig JW, Teeuwen C, Vermeulen FEE, Welten RJT. Benefits of carotid patching: a randomized study. J Vase Surg 1988;7:240-7. Ranaboldo CJ, Barros D'Sa AAB, Bell PRF, Chant ADB, Ferry PM. Randomized controlled trial of patch angioplasty for carotid endarterectomy. Br J Surg 1993;80:1528-30. Clagett GP, Patterson CB, Fischer DF Jr, Fry RE, Eidt JF, Humble TH, Fry WJ. Vein patch versus primary closure for carotid endarterectomy: A randomized prospective study in a selected group of patients. J Vase Surg 1989;9:213-23. Cuming R, Blair SD, Powell JT, Greenhalgh RM. The use of duplex scanning to diagnose perioperative carotid occlusions. Eur J Vase Surg 1994;8:143-7. Golledge J, Cuming R, Ellis M, Beattie DK, Davies AH, Greenhalgh RM. Clinical follow-up rather than duplex surveillance after carotid endarterectomy. J Vase Surg 1996;25:55-63. Ellis M, Cuming R, Laing S, Vashisht R, Franks PJ, Greenhalgh RM, et al. The reproducibility of colour-coded duplex scanning in measuring arterial wall dimensions. Eur J Vase Endovasc Surg 1992;6:386-9. Baker WH, Koustas G, Burke K, Littooy FN, Greisler HP. Intraoperative duplex scanning and late carotid artery stenosis. J Vase Sm'g 1994;19:829-33. Kinney EV, Seabrook GR, Kinney LY, Bandyk DF, Towne JB. The importance of intraoperative detection of residual flow abnormalities after carotid endarterectomy. I Vase Surg

1993;17:912-23. 18. Barnes RW, Nix ML, Nichols BT, Wingo JP. Recurrent versus residual carotid stenosis. Ann Surg 1986;203:652- 60. 19. Sanders EACM, Hoeneveld H, Eikelboom BC, Ludwig JW, Vermeulen FEE, Ackerstaff RGA. Residual lesions and early recurrent stenosis after carotid endarterectomy: a serial follow-up study with duplex scanning and intravenous digital subtraction angiography, l Vase Surg 1987;5:731-7. 20. SalvianA, Baker JD, Machleder HI, Busuttil RW, Barker WF, Moore WS. Cause and noninvasive detection of restenosis after carotid endarterectomy. Am J Surg 1983;146:29-34. 21. Colyvas N, Rapp JH, Phillips NR, Stoney R, Perez S, Kane JP, Havel RJ. Relation of plasma lipid and apoprotein levels to progressive intimal hyperplasia after carotid endarterectomy. Circulation 1992;85:1286-92. 22. Ohja H. Wall shear stress temporal gradient and anastomotic intimal hyperplasia. Circ Res 1994;74:1227-31. 23. Farm JI, Sokoloff MH, Sarris GE, Yun KL, Kosek JC, Miller DC. The reversibility of canine vein-graft arterialization. Circulation 1990;82(5 suppI):IV9-18. 24. Dobrin PB, Littooy FN, Endean ED. Mechanical factors predisposing to intimal hyperplasia and medial thickening in autogenous vein grafts. Surgery 1989;105:393-400. 25. Thomas M, Otis SM, Rush M, ZyroffJ, Dilley RB, Bemstein EF. Recurrent carotid artery stenosis following endarterectomy. Ann Surg 1984;200:74-9. 26. DeLetter JA, Moll FL, Welten RJ, Eikelboom BC, Ackerstaff RG, Vermeulen FE, Algra A. Benefits of carotid patching: a prospective randomised study with long-term follow-up. Ann Vase Surg 1994;8(1):54-8.

Submitted Sep. 3, 1996; accepted Nov. 25, 1996.